Friday, August 28, 2015

Fall Antenna Plans: 80 Meters

Now that the 80 meter half sloper is reconnected (I had borrowed the coax for the 6 meter yagi over the summer) I am reminded how awful an antenna it really is. This is a good time to consider alternatives since the noisy summer conditions are still prevailing on the low bands, and will continue for a few more weeks.

To summarize, here are the problems I have with my half sloper:
  • The antenna has a significant horizontally-polarized component in many directions. Despite the use of the tower as a major component of the antenna it seems that the wire can often dominate the far-field pattern. Much of the antenna's radiation is at high angles, which is not what I want.
  • There are substantial ground losses which severely cut into its efficiency. I am comfortable claiming this based on its on-air performance, feed point impedance (presence of series ground loss resistance) and the software model. NEC2 often has been reported to underestimate losses from real ground, yet even so the modelled losses are substantial.
  • Antenna bandwidth is modest. It is cut to resonance within the CW segment (3.5 to 3.6 MHz) and does poorly in the SSB segment (3.8 MHz). I've tried the FT-1000 MP's ATU and an external antenna tuner with poor results. I can match the antenna for SSB, but the matching and SWR loss seem quite high. While adjusting feed line length might help, I'd rather have a broadband match.
These are problems whether I am contesting with QRP or DXing with 100 watts on 80 meters. Both have been frustrating. I had higher hopes for the half sloper. In retrospect I feel my initial optimism was misplaced. Now it's time to do something about it.

Options

As I move lower in frequency my options become increasingly constrained by my available supports and property. Just moving a factor of 2, from 40 down to 80 meters, I can pretty much rule out all horizontal antennas. They will either do no better than what I have or will destructively interact with may high band antennas. Interaction is unacceptable since 80, for now at least, is less valuable for QRP contesting and for DXing.

Regardless of the antenna I choose, horizontal or vertical, ground losses must be addressed. The ground directly under all my antenna is poor, not the medium ground I typically use in my models. I model that way to make the results meaningful for most readers of this blog. My lot is backfilled with sand (septic tile bed) over shale. Soil is only 12" (30 cm) deep.

Do not be deceived by advertising for commercial vertical antennas with "no radials" designs and that have elaborate matching networks, or any home-built antenna that does not require radials, such as a delta loop or half sloper. You cannot so easily dismiss ground interactions, and the inevitable loss.

Radials are needed, although I don't want them. There is too much traffic in my backyard to have wire lying on the surface, even if it is worked into the grass and weeds. Burial is out of the question at this QTH since it would tear up the lawn for what is likely to be a brief deployment. A vertical would have to use the tower that is centred in the yard since it is the only location where radials of any kind can be placed.

As I said, my options are limited. Whether I like it or not I have to go vertical and find a way to put down some radials. This is the only way I can hope to improve low-angle radiation and reduce ground loss. Reducing local noise (QRN), which can be quite severe during the evening, is out of the question since I have no space for a directive receive antenna. With QRP that's rarely a problem since I am far more concerned with being heard. A separate, directive receive antenna is of little value for the same reason.

My objectives for the 80 meter antenna:
  • Low-angle radiation suitable for DX. I only need enough high-angle radiation to make contest contacts with the nearby US northeast and midwest.
  • The minimum radial field to reduce ground loss to an acceptable level.
  • Feed system to achieve a 50 Ω match from 3.5 to 3.8 MHz, for the tower, top loaded with a tri-band yagi, and assorted cable runs.
With those points in mind we are ready to proceed.

Resonance

There are so many variables to determining the resonant frequency of a vertical constructed from a tower and yagis that it is often best to just go out and measure it. Unfortunately this requires that the radial system be in place. For a small number and length of radials the radials play a substantial role in determining the resonant frequency. With more and longer (λ/4) radials the radials tend toward non-resonance, leaving the monopole itself as the principal tuning variable.

Calculation can get us close, or at least close enough to guide the design of the radial system and feed network for a tower with a yagi on top. There is a formula for this in the 1st edition of ON4UN's Low-band DXing, page II-32:

L = 0.38f ( H + SQRT( S ( 1000 - H ) / 500 )

L is the electrical length in degrees (λ/4 = 90°), H is the height in feet, S is the area of the yagi in square feet, and f is the frequency in MHz. This formula is not in the 5th edition, having been replaced by a graph for a more restricted range of figures. My guess is that this was done to discourage readers from applying the formula to extreme cases where it is inaccurate. Therefore only use the formula as a rough estimate.

Per this formula, my tower plus mast height of 15 meters, with an Explorer 14 at the top, has an approximate electrical length of 105° at 3.65 MHz. It should therefore be resonant below the band edge. This is close enough to λ/4 to allow a simple matching network. I like that.

However there is more involved in getting a good match on a ground mounted vertical. With a small quantity of radials the radial length contributes to the resonant frequency. With more and longer ones the radial system becomes non-resonant. I will have a few short ones, so it matters.

Ground loss

There are ample resources in the amateur literature about ground loss and the mitigation of ground loss. One in particular I like is the 5th edition of ON4UN's book. On the internet, one good place to look is the series of articles by N6LF. Read them if you have or intend to build a vertically-polarized antenna for the low bands. I will only say a few words here that are specific to my situation on 80 meters.

Z = ( Rrad + Rgnd + Rant ) + jX

The resistance term R in the antenna feed point impedance Z is composed of 3 series resistances:
  • Rrad: Radiation resistance
  • Rgnd: Ground loss
  • Rant: Conductor and ESR (equivalent series resistance) loss in the antenna and matching network
For a typical λ/4 vertical monopole with a perfect ground plane the radiation resistance is 37 Ω. Ground loss due to poor ground or ground plane can be larger than the radiation resistance. Conductor loss is typically negligible for an antenna of this type, however capacitors, inductors, transformers and transmission line stubs in the matching network (especially for short verticals) can also be quite large.

The only way to manage loss outside of the near field (far field ground reflections) is to move to a better QTH! This is outside of my control for the time being, as it is for most hams. What I hope to manage is near field ground loss close to the antenna. This is where the radials come into play.

Using EZNEC, I built a model of a vertical with two grounds.The first ground, out to 8 meters distance from the tower, is poor (0.002, 10) to account for the sand fill and bedrock. From 8 meters outward I use a medium ground (0.005, 13). The intention is to approximate near-field ground losses without distorting the far-field pattern. That's the best I can do, knowing that nearby buildings are within the near field. But then that's true for any low band antenna on my property. There is a significant change in ground loss when the media parameters for that inner ground are varied.

Matching network components should be selected for their low ESR. It is a mistake to only pay attention to the component values and their maximum voltage and current ratings. Transmitting capacitors (fixed or variables) and high-Q coils are best. The loss is there at all power levels so don't take shortcuts if you, like me, operate QRP.

Radial system

The radial system is unlikely have more than 8 radials. Their maximum length is limited by the size of my backyard and fixtures. Toward the south I can go quite long. North toward the house allows lengths up to 15 meters, and shorter where I have the deck and and landscaping. The main limitation is east and west, where the maximum possible length is 7 meters. Although my lot is ¼ acre it is only 15 meters (50') wide.


With this constraint my choices are to go long where I can and short where I cannot, or to choose equal but short lengths for all radials. The question is which does better? Radial asymmetry is something I've dealt with before, its good and bad points. Once again, EZNEC helps to answer this question.

In the view at right (with currents plotted), X is east and Y is north. My lot is 15 meters wide along the X axis. My house is about 15 meters north of the tower. Going south, there is about 25 meters of available space.

In this first model wires #2 and #4 are 7 meters long. The diagonal wires are 10 meters long and wires #3 and #5 are 15 meters long. It turns out this is a poor arrangement.

Most of the radial current is in the 6 longest radials, with almost no current in the 2 short ones. When the current is so low their effectiveness is quite poor. Snipping those two wires from the models had a negligible impact on ground loss (-0.1 db). The azimuth pattern is omni-directional for all reasonable arrangements of 6 or 8 radials. It is just ground loss and resonant frequency that are effected.

When the two longest radials are reduced to 10 meters length the current becomes more equalized and ground loss is slightly reduced. The relative current flowing in the 7 meter long radials is also higher. I decided to proceed through the rest of the modelling with this arrangement.

Antenna feed and matching network

For the typical shunt-fed tower it is necessary to use a gamma or omega match since the tower is not isolated from ground. The combination of electrical continuity into the concrete-embedded tower base comprises a Ufer ground. My tower is isolated from ground, though not in an ideal fashion. The preserved wood base is a poor conductor, even when wet, and its ground contact area is less than 1 square meter.

It may be worth an experiment to feed the tower directly. If the feed point impedance at resonance comes reasonably close to 50 Ω I'll take it as an indication that the tower's ground isolation is acceptable (see below).

To begin the analysis I added a radial system to a vertical monopole in EZNEC. The radial system has 8 radials of 10 meters length, except for the east and west radials which can only be 7 meters long (see discussion above). Since the radials in this configuration affect system resonance, the monopole must be tuned once the radial lengths are set.



The SWR plot above has the monopole adjusted to an electrical length that is resonant at 3.6 MHz.Notice that the feed point impedance is 60 Ω. This is 23 Ω higher than the 37 Ω of a ground-mounted vertical with an ideal (zero loss) radial system. Assuming this reflects reality (which is unlikely) the near field ground loss would be approximately -2 db. This does not include other environmental loss, such as nearby houses and ground reflections beyond the near field. The true ground loss almost certainly will be greater.

Ground loss is undesirable. Yet if it can't be avoided we can at least use it to our advantage. As the cliche goes, when life serves us lemons, make lemonade! Here we find that the ground loss makes it possible to achieve a broadband match to 50 Ω coax without a matching network. This is not unlike some commercial tri-band yagis where the trap loss permits a direct match to 50 Ω coax even though the radiation resistance may be half that value.

All we now must do is add a series capacitor (between the coax centre conductor and the tower) to compensate for the inductive reactance due to the tower's resonance at a lower frequency. The capacitor would be adjusted for minimum SWR at the selected centre frequency of 3.6 MHz. Tuning can be done with a variable capacitor, and then substituting a fixed capacitor of the required value.

Should direct feed result in a higher SWR (due to actual ground loss or excessive ground interaction at the tower base), a gamma match is the next best bet. The estimated requirement is a gamma rod (or wire) about 7 meters long, a series capacitor to tune the gamma match and tying the radial system to the tower base.

In the unlikely event this is insufficient, an omega match would be required. Were I to attempt to shunt feed the tower on 160 meters, an omega match would certainly be required since the electrical length of the tower would only be 53° at 1.85 MHz. The additional capacitor would add loss, though it would be small in comparison to higher ground loss of the small radial system.

Transmission line

In an ideal vertical installation the coax transmission would be buried to minimize antenna coupling, and would have a substantial common mode choke at the base of the vertical. My situation is far from the ideal, and that may be okay. The coax will have to run overhead, in parallel with all my other cables out to the tower, to avoid damage to the cable and to people.

Some amount of coupling and therefore common mode current is going to be unavoidable. Even were I to follow the ideal for this one antenna, there would still be coupling to the several cables running up the tower to the rotator and other antennas. I have that very situation today with the loaded half sloper antenna for 80. There is coupling, though not enough to be a problem. Running a kilowatt would change my opinion, but that won't happen.

In all likelihood I will run the coax overhead, with all the other runs, and have it dip down toward the feed point at the tower base. Since I will be using a 130' (40 meter) length of RG-213 there will be enough spare coax to wind a coaxial choke. That choke must not be "scramble wound" because the inter-turn capacitance would render the choke ineffective. However, I am unconvinced that a properly wound coax choke is worth the trouble because I know there will be coupling back to the shack on the other cables. This decision will be deferred until I build the antenna.

The plan

I'll be removing the half sloper when I am ready to install the second 40 meter inverted vee. I plan to do that in the next two weeks (early September). When that is done, and the septic tank is pumped, I'll be ready to proceed with the 80 meter vertical. Allowing time for experimentation and a further delay toward the end of lawn mowing season, I should have radials in place and the antenna ready by early October.

Elevation pattern of my tower vertical with the ground,
radial system and feed system discussed in this article
One problem I foresee is the ability to confirm performance with a reference antenna. Unfortunately the half sloper cannot be kept as that reference. Using a 40 meter antenna with a tuner is a poor alternative since, in earlier testing with this arrangement, its performance is very poor and therefore unsuitable as a reference. I have no good third alternative.

Based on models alone, it appears I can expect at least 3 db low-angle gain improvement, and probably more, in comparison to the half sloper. Some of that comes from the switch to vertical polarization and the rest from reduction of near-field ground loss by using radials, poor as those radials must be.

I may have to roll up the radials when not in use, at least until early November when people traffic in the yard is no longer a factor. Then it'll be safe until spring thaw.

Once the antenna is built, tuned and I have some performance observations from on-air use I will follow up on this article. You can then compare my modelling alternatives with my final choice of feed and radials.

Sunday, August 16, 2015

Fall Antenna Plans: 40 Meters

Fall is rapidly approaching. For me that means preparing the antennas for fall and winter contests, and the occasional DXpedition. Summer has been a down time for my radio activities, with travel, home improvements, sports and other activities. The articles in this blog have been sparse. Now is the time to get serious about antennas.

The 6 meter yagi is now out of the way, making room for other antennas. There is little I can do for the high bands (20 through 10 meters) with my current supports so there will be no changes for those antennas. It is 40 and 80, and possibly 160, where I desperately need improvement. This is especially true if I again enter the QRP category in the major contests. On these bands every decibel counts when others can barely copy me, or not at all.

Final decisions have yet to be made. Options are limited, as are the performance improvements. I'll take you through my thinking at this stage so you'll understand my choices. Later, when the antennas are built and tested, I'll relate how I came to do what I will have done. In this article I'll discuss 40 meters, deferring the other bands to future articles.

Pattern

My only antenna for 40 at present is a multi-band inverted vee at an apex height of 14 meters. One end is tied to the tower and the other to the house eaves. It is asymmetrical, though not by a lot. It is enough to skew the pattern, as you can see in the adjacent plot.

An inverted vee is not really an omnidirectional antenna, even if it is often billed as such. It is only more so than a dipole. You can read how I came to choose this antenna in an article I wrote in 2014.

Notice the azimuth pattern, at a DX optimum elevation angle of 10°. I set up the EZNEC model so that east is to the right. The antenna is replete with compromises.

Europe is down -2 db from where it peaks, as is much of the US. Asia is down -5 db and the southern US, Caribbean and South America are even worse. My operating experience tells me quite clearly this is hurting my contest scores and DX performance. I need a better or a second antenna to fill these gaps.

My best options are as follows, as constrained by my property and supports. There will be no new tower at this QTH this year, and perhaps never.
  • Inverted vee: Mounted at the top of the tower this antenna would have the same 14 meter apex height as the multi-band inverted vee. Made symmetric with an interior angle of 90° it would be oriented to be at approximately a right angle to the other inverted vee. In this way interaction with the tri-band yagi on 15 meters above it and the vees would be small. While a tight squeeze I have tie off points selected that would make this work.
  • Omega-tuned boom dipole: This involves extending the boom of the Explorer 14 by at least 6 meters and making an omega match with the tap being a wire tied to the end of the existing boom and angled downward to the mast just above the mast bearing. This is a tried and true design that I previously considered but rejected because of the complexity of raising the extended yagi. That remains a task I am loathe to tackle.
  • Rotatable dipole: The driven element of my recently-purchased Cushcraft XM240 can be mounted alone on the mast and rotated. It would be mounted parallel to the yagi's boom to avoid interaction, which I've modelled and seen that it is quite severe on 15 meters though acceptable on 20 and 10. At 43' long (13.5 meters) it fits within my 50' wide lot but in windy conditions could tangle with the trees that serve as the tower's guy anchors. Some careful measurement would be required. The wind load is substantial (~3 ft²), so it would have to come down after the winter season, before spring and summer storms arrive.
All of these antennas would have a feed point at the top of the 14 meter tower (DMX-52) that currently supports a Hy-Gain Explorer 14. The RG-213 transmission line is what was used for the 80 meter half sloper and was temporarily used for the 6 meter yagi. That coax is now free since I have other plans for 80.

I modelled the inverted vee and rotatable dipole in EZNEC and overlaid them on the azimuth pattern of the existing inverted vee. All are at an elevation angle of 10°.

First, notice that the inverted vee on the tower has less gain at low angles. This is because the interior angle is smaller than on the multi-band inverted vee. The difference is about -1 db. Due to that and constraints on tie off points, it does only a modest job of filling in those pattern gaps. Northwest and southeast are worst.To the north the improvement is just 2.5 db and to Europe there is no difference. Where the new antenna does well is to the south where it is ~6 db better. This pattern of this inverted vee is more omnidirectional due to its symmetry, a symmetry that might not survive real-world interactions.

As expected the XM240-derived dipole does best, since its average height is higher than the inverted vees. Even with the coil loss (-0.36 db) the dipole gain peaks 0.5 db better than the multi-band inverted vee. (NB: in an earlier article we saw that a dipole typically beats an inverted vee with a 120° interior angle by 1 db.)

The performance improvement comes at the expense of wind load and the need to rotate the dipole to get the most from it. Of course with both antennas online it is possible to switch instantly. At times when both high and low bands are open there is the additional matter of the dipole pointing of the side of the tri-band yagi. But it does allow working the US and points north and south while the yagi is pointed to Europe.

Interaction

The model view at right, with currents, shows the high coupling when the dipole is rotated to where both antennas are in the same plane. It turns out that interactions are a major determining factor as to which antenna I will go with. This can, and does, degrade both antennas' performance.

The nearness of the ends of the dipole and inverted vee is not due to perspective; they are ~3 meters apart (the vee is tied off to the tower). For model simplicity the other elements of the multi-band vee are omitted since their effect is negligible.

To test this out in the EZNEC model I fed each antenna in turn to estimate the effect of interaction. I then rotated the dipole to measure how the interaction changes with direction.

Interactions when the inverted vee and dipole are in the same plane (left) and orthogonal (right), at 10° elevation
When orthogonal there is little interaction. Compare the right pattern with the one at the top of this article. (The vee's pattern is not skewed; it has been rotated in the interaction model.) Low current on the inactive antenna confirms this. When collinear (in the same plane) the impact is large. Antenna lobes are no longer where we want them! However there is a gain effect, as in any parasitic array, with the amplitude of the major lobes 1 to 2 db higher than in the non-interacting (orthogonal) orientation. Unfortunately this gain is not useful or reliable.

The SWR for both antennas in the collinear case are poor. Resonance is shifted and bandwidth is reduced for both antennas. Even if it were possible to live with the pattern distortion, the large swing in SWR is not tenable for most operating, especially contests, and would at least require frequent adjustment with an antenna tuner (preferably automatic).

Where I go from here

This analysis demonstrates the importance of checking interactions between antennas for the same or harmonically-related bands. My antenna decision is heavily constrained by interactions. I could not put up a rotatable dipole unless I remove the 40 meter element from the inverted vee. That is not desirable.

In my earlier interaction modelling I was able to demonstrate modest interaction on 15 meters between the tri-band yagi and the 40 meter inverted vee. Again, worst case was when they were collinear (yagi pointed east). I judged the degradation to be acceptable: about -1 db gain, poorer F/B but little effect on SWR. Compromises are sometimes unavoidable. But you can't make an informed decision until the interactions are tested, in a model (preferred) or in the field. Spending time on this analysis saved me time and effort, and disappointment.

My choices now come down to these options:
  • Install the rotatable dipole and remove the 40 meter inverted vee. I would lose instant direction switching and have to do some mechanical work to modify the inverted vee and to install the dipole on the tower. The higher wind load is a risk, though I believe it is managable if I take the dipole down in early spring. But then I would have no 40 meter antenna at all.
  • Build and install a second, 40 meter only, inverted vee on the tower, orthogonal to the first. I would retain instant direction switching and avoid destructive interactions. The 90° interior angle of the vee reduces gain but is necessary to avoid degrading the yagi's performance on 15 meters. Installation of this vee presents a mechanical challenge in that one leg has to be tied off at some height on a suitably placed tree. There is a wasp nest in the way (I'm allergic) that I'll have to first remove!
I'll report back on what I decide to do. I may come up with a further option to consider.

A subsequent article I will discuss my options for 80 and 160. I have fewer good options for those bands. I may decide to entirely forgo 160 this season.